Scalable service-oriented replication with flexible consistency guarantee in the cloud

Abstract

a b s t r a c tReplication techniques are widely applied in and for cloud to improve scalability and availability. In such context, the well-understood problem is how to guarantee consistency amongst different replicas and govern the trade-off between consistency and scalability requirements. Such requirements are often related to specific services and can vary considerably in the cloud. However, a major drawback of existing service-oriented replication approaches is that they only allow either restricted consistency or none at all. Consequently, service-oriented systems based on such replication techniques may violate consistency requirements or not scale well. In this paper, we present a Scalable Service Oriented Replication (SSOR) solution, a middleware that is capable of satisfying applications' consistency requirements when replicating cloud-based services. We introduce new formalism for describing services in service-oriented replication. We propose the notion of consistency regions and relevant service oriented requirements policies, by which trading between consistency and scalability requirements can be handled within regions. We solve the associated sub-problem of atomic broadcasting by introducing a Multi-fixed Sequencers Protocol (MSP), which is a requirements aware variation of the traditional fixed sequencer approach. We also present a Region-based Election Protocol (REP) that elastically balances the workload amongst sequencers. Finally, we experimentally evaluate our approach under different loads, to show that the proposed approach achieves better scalability with more flexible consistency constraints when compared with the state-of-the-art replication technique.

A brief survey on replica consistency in cloud environments

“…Similarly, Indigo [12] does not guarantee availability, thus achieving consistency and partition tolerance. Finally, SSOR [22] does not guarantee availability and partition tolerance, which means that its method concerns only in providing consistency.…”

“…Thus, in order to improve scalability a relaxed consistency state is often provided by some systems. However, the price is that the state of each replica may not be always the same [22].…”

High availability in clouds: systematic review and research challenges

“…The main problem associated with this service is the question of how to govern the trade-off between consistency and resource usage [18]. In Clouds, the replication may be achieved either by copying the state of a system (checkpoint) or by replaying input to all replicas (lock-step based) [16] (see Fig.…”

Toward Formal Modeling and Verification of Resource Provisioning as a Service in Cloud

“…If client, service manager and resource service do not reach consistent states, there may be resources misused or even abandoned. Many researchers have studied consistency problems in cloud services, such as data consistency [3], storage consistency [4], [5], queuing service consistency [6] and replication consistency [7], [8]. However, they do not consider resource provisioning from the perspective of service, and are rarely concerned about consistency in the process of resource delivery.…”